تأمین زیان-وام بانک، قوانین بانک مرکزی در مقابل برآورد : مورد پرتغال

A fair level of provisions on bad and doubtful loans is an essential input in mark-to-market accounting, and in the calculation of bank profitability, capital and solvency. Loan-loss provisioning is directly related to estimates of loan-loss given default (LGD). A literature on LGD on bank loans is developing but, surprisingly, it has not been exploited to address, at the micro level, the issue of provisioning at the time of default, and after the default date. For example, in Portugal, the central bank imposes a mandatory provisioning schedule based on the time period since a loan is declared ‘non-performing’. The dynamic schedule is ‘ad hoc’, not based on empirical studies. The purpose of the paper is to present an empirical methodology to calculate a fair level of loan-loss provisions, at the time of default and after the default date. To illustrate, a dynamic provisioning schedule is estimated with micro-data provided by a Portuguese bank on recoveries on non-performing loans. This schedule is then compared to the regulatory provisioning schedule imposed by the central bank.

Fair provisioning on bad and doubtful loans is of great importance for investors and bank regulators. Consider the merits of Basel II, the revised capital accord that would much better capture the actual risks taken by banks (Basel Committee, 2004). It is quite evident that this accord will not have much relevance if the measurement of bank capital is not satisfactory. A key input in the measurement of bank capital is the amount of loan-loss provisions1 on bad and doubtful loans. Well-known cases of significant under-provisioning in recent history include the French Credit Lyonnais in 1993, Thailand in 1997, Japan in late 1990s (Genay, 1998), and, more recently, China. A fair level of loan-loss provisions is needed to measure bank profitability, capital, and solvency. The issue of adequate loan-loss provisions is well recognized by central banks. In Basel II, the difference between provisions and expected loss on the loan portfolio will affect the measure of capital. In Portugal for example, the central bank has imposed a provisioning schedule on non-performing loans. Provisions increase as time elapses since a loan was declared ‘non-performing’. However, the dynamic provisioning schedule is ‘ad hoc’, not based on empirical studies. Empirically-based measures of provisions on non-performing loans are much needed. As bank loans are, by their economic nature, private, there is not much market-based information to assess their current value at the time of distress in many countries, so that loan-loss provisions must often be estimated. Loan-loss provisioning is directly related to estimates of loan-loss given default (LGD). A literature on LGD on bank loans is developing but, surprisingly, it has not been exploited to address, at the micro level, the issue of provisioning at the time of default, and after the default date. As the likelihood of being repaid diminishes as time elapses after the default date, a dynamic schedule of provisioning is needed. In this study, we build on a recent paper (Dermine and Neto de Carvalho, 2006) to show how to estimate a dynamic provisioning schedule. To illustrate the methodology, micro data provided by a large Portuguese bank on recoveries on non-performing loans are used to estimate a dynamic provisioning schedule, which is then compared to the one imposed by the Central Bank of Portugal. The paper is structured as follows. The literature on bank loan-loss provisioning is reviewed in Section 2 of the paper. The provisioning schedule of the Central Bank of Portugal is presented in Section 3. The mortality-based approach to analyzing fair provisioning on bad and doubtful loans is discussed in Section 4. The data set is presented in Section 5. Empirical univariate estimates of dynamic provisioning are presented in Section 6, and these estimates are compared to the schedule of the Bank of Portugal. Finally, a multivariate statistics approach to loan-loss provisioning is developed in Section 7. Section 8 concludes the paper.

A fair level of provisions on bad and doubtful debt is an essential part of capital regulation and bank solvency. In Portugal, for example, banks have to apply a mandatory provisioning schedule imposed by the central bank. The purpose of the paper is to show how micro data on recovery overtime on bad and doubtful loans allows computing dynamic provisions, at the time of default and over time. A first univariate mortality-based approach allows us to compute provisions for three classes of loans: no guarantee/collateral, personal guarantee only, and collateral with or without guarantee. The mortality-based approach allows using the entire dataset. The reason is that cumulative recovery rates are calculated from marginal recovery rates which can be computed on all loans, including recent distressed loans with short recovery history. A second multivariate approach facilitates analyzing more precisely the determinants of loan recoveries and provisions over time. A drawback of this approach is that one can only use loan data with a long recovery history. The two methodologies are applied to a data set on recoveries on bad and doubtful loans provided by a large Portuguese bank. Three main results are as follows. First, bad and doubtful loans with no guarantee/collateral exhibit better recoveries than loans with personal guarantee. This could be due to the fact that the decision to lend without guarantee took into account the higher expected recovery rates. Second, the past recovery history has a highly significant positive impact on future recovery. Third, a comparison with the Bank of Portugal mandatory provisioning rules indicates some regulatory conservatism in calling for 100% provision 31 months after the default date, when, in fact, significant amounts are still recovered after that date. But, much more stringent provisions should be enforced in the shorter run, before the “90-day”—provisioning trigger date. A word of caution is that the empirical results, being based on a dataset of a single bank in a specific time period, can capture some of the bank's idiosyncrasies. Additional empirical studies are needed to validate the empirical findings of the paper, but the two methodologies presented in this paper provides a basis to estimate loan-loss provisions on bad and doubtful bank loans.